News for hackers. Stuff that matters.
Re:Constitutional AI: RLHF on Steroids by eterevsky on Tuesday May 09, 2023 @04:47AM (#35871669)
"Accurate" is a component in "helpful".
Re:Constitutional AI: RLHF on Steroids by BryantD on Tuesday May 09, 2023 @10:58AM (#35875329)
Indeed. It could also be a component in "ethical," depending on what your ethics are.
Re:Constitutional AI: RLHF on Steroids by cubefox on Tuesday May 09, 2023 @09:02PM (#35882371)
No it isn't. Harmlessness, helpfulness, honesty are separate components. See paper. A model could be very helpful and tell you everything you want to hear, but at a cost of engaging in deception. E.g. in regards to politically incorrect information.